Predatorprey Interactions Between the South Polar Skua Catharacta

Home , South polar skua

Journal of Avian Biology 44: 089–095, 2013 doi: 10.1111/j.1600-048X.2012.05731.x © 2013 The Authors. Journal of Avian Biology © 2013 Nordic Society Oikos Subject Editor: Thomas Alerstam. Accepted 31 October 2012

Predator–prey interactions between the South Polar skua Catharacta maccormicki and Antarctic tern Sterna vittata

Karel Weidinger and Václav Pavel

K. Weidinger ([email protected]) and V. Pavel, Dept of Zoology and Laboratory of Ornithology, Faculty of Science, Palacky Univ., Tr Svobody 26, CZ-771 46 Olomouc, Czech Republic.

Antarctic terns have to co-exist in a limited space with their major nest predator, the skuas. We conducted artificial nest experiments to evaluate the roles of parental activity, nest location and nest and egg crypsis in this simple predator–prey system. Predation on artificial (inactive) nests was higher in traditional nesting sites than in sites previously not occupied by terns, which suggests that skuas memorized past tern breeding sites. Predation on artificial nests in inactive colonies was higher than in active (defended) colonies. Parental defense reduced predation in colonies to the level observed in artificial nests placed away from colonies. This suggests that communal defense can balance the costs of attracting predators to active colonies. Within colonies, predation was marginally higher on experimental eggs put in real nests than on bare ground. Although it seems that the presence of a nest is costly in terms of increased predation, reductions in nest size might be constrained by the need for protective nest structures and/or balanced by opposing selection on nest size. Predation did not differ markedly between artificial (quail) and real tern eggs. A simultaneous prey choice experiment showed that the observed predation rates reflected egg/nest detectability, rather than discrimination of egg types. In sum- mary, nesting terns probably cannot avoid being detected, and they cannot defend their nest by attending them. Yet, by temporarily leaving the nest, they can defend it through communal predator mobbing, and at the same time, they can benefit from crypsis of unattended nest and eggs.

The role of predation risk in the evolution and/or However, from the perspective of less abundant prey species, maintenance of avian coloniality is still controversial the impact of predation can be more detrimental than for (Wittenberger and Hunt 1985, Siegel-Causey and more abundant and/or large species, where predation is Kharitonov 1990, Clode 1993, Kenyon et al. 2007, Sachs diluted by prey numbers (Brooke et al. 1999, Votier et al. et al. 2007, Varela et al. 2007). Nevertheless, currently 2004) and may be further reduced through nest defense observed nest predation patterns often suggest an adaptive (penguins; Young 1994) or selection of protected nest sites significance of colonial nesting in terms of increased breed- (some petrels; Varpe and Tveraa 2005). ing success. Seabirds, because of their prevalent habit of The Antarctic tern Sterna vittata is a circumpolar species group living, are a traditional model in studies of costs and breeding on sub-Antarctic islands and the Antarctic benefits associated with colonial breeding (Clode 1993). Peninsula to 68°S, in loose colonies containing from a few to Seabirds inhabiting species-poor polar areas offer an addi- hundreds of pairs. Both eggs and chicks suffer from heavy tional advantage of being relatively simple predator–prey avian predation and inclement weather, the overall nest systems, in particular species breeding in the Antarctic area, success being generally low and highly variable among where they have never been exposed to terrestrial predators. geographical locations, years, as well as between colonies The largest avian breeding aggregations in the Antarctic within a year (Kaiser et al. 1988, Parmelee 1992, Jabłoński are formed by penguins and petrels, whose eggs and chicks 1995, Higgins and Davies 1996, Casaux et al. 2008). Here, are the major terrestrial prey available to the local avian we report on nest predation in the Antarctic tern at the predators – skuas (Catharacta spp.; Parmelee 1992, Young southern border of its breeding range, where the species 1994, Weidinger 1998, Brooke et al. 1999, Van Franeker breeds close to its ecological limits. Terns co-exist here in a et al. 2001). Yet, the comparatively smaller colonies of limited space with their major avian predator, the South other Antarctic flying seabirds have received little attention Polar skua Catharacta maccormicki. in this respect (Parmelee and Maxson 1974, Kaiser et al. We conducted artificial nest experiments to evaluate 1988), possibly because they are – from the perspective of the respective roles of nest site location, nest and egg crypsis the predator (and also researcher) – a negligible prey source and parental activity in the predator–prey interaction (for a review of skua diet see Higgins and Davies 1996). between skuas and terns. Nesting terns probably cannot

89 avoid detection by skuas because of their conspicuous behav- with Japanese quail Coturnix japonica eggs as a surrogate ior and the good visibility of incubating birds against the for real tern eggs. The quail eggs were hardboiled and X-ray nest background. Yet, frequent changing of nesting sites sterilized just before transportation to Antarctica, to prevent between years (Parmelee 1992, Peter et al. 2008) and possi- pathogen transmission. These procedures did not influence bly re-nesting at different sites within a year (Parmelee 1992) the external appearance of the eggs. Quail eggs are cryptic suggest some benefit from the abandonment of previous (to the human eye), but smaller than tern eggs (32  25 vs breeding sites. This is to be expected if skuas remember and 46  33 mm). In addition to quail eggs, we also used a lim- revisit tern nesting sites once they are discovered. We asked ited number of real tern eggs that we collected from nests 1) whether artificial nests – without parental activity that that were found already abandoned by the time of our may disclose them to skuas – would be preyed upon at a arrival to the study area. All artificial nests contained one higher rate if they are located in inactive tern colonies egg, completed natural tern clutches contained on average compared to sites not previously occupied by nesting terns. 1.2 eggs (Weidinger and Pavel in press). We expected 2) that this effect, if present, would increase We defined the following experimental treatments with distance between the colony and the unoccupied site. (Appendix 1). (T1) un-manipulated active tern nests. (T2) Contrary to attracting predators’ attention, breeding quail eggs placed in already inactive real tern nests in aban- terns can actively defend their nests by communally doned colonies. We used all discovered current-year nests mobbing predators to drive them away from the colony within a particular colony/nest cluster, regardless the size (Parmelee and Maxson 1974, Jabłoński 1995, Whittam and of nest structure. (T3) quail eggs placed on a suitable sub- Leonard 2000). We thus also asked 3) whether artificial nests strate within inactive colonies. The sites were chosen as to in inactive colonies – without parental defense – would form pairs with T2 nests, separated from each other by a experience higher predation rates than defended nests in distance of 10 m. (T4, T5) tern and quail eggs, respectively, active colonies. Because incubating parents temporarily placed on suitable substrate in an alternating sequence at leave their nest to participate in mobbing, the attacking 10 m intervals; the transects were located  100 m from predator has to search for individual unattended nests the edge of inactive colonies. (T6) quail eggs placed on suit- within a colony. In such cases, cryptic eggs would offer a able substrate in transect at 10 m intervals; the transects selective advantage (Blanco and Bertellotti 2002, Sanchez were located  1 km from the nearest tern breeding sites et al. 2004). Because egg coloration may be selectively neu- (active or abandoned) and about 50 m from an active skua tral if predators detect nests before detecting eggs (Götmark nest. Deployment of artificial nests (T4, T5, T6) in transects 1993), we first asked 4) whether within a colony, experi was dictated by logistical constraints. Although this spatial mental eggs placed in real nests would experience higher pre- pattern differs from that of natural nests, we believe this dation rate than eggs placed on bare substrate. Finally, did not influence foraging behaviour of skuas to a degree we asked 5) whether experimental eggs (presumably less that might invalidate the results. We suppose that skuas cryptic) would experience higher predation rates than real detect artificial nests one at a time during overflights, not by tern eggs, if both egg types are exposed on bare substrate, to walking between neigbouring nests. The 10 m interval was exclude any effects of the nest per se. chosen as to simulate the inter-nest distance in the core areas of real colonies (for account of nesting densities see Kaiser 1996 and Weidinger and Pavel in press). Material and methods The artificial nest treatments T2–T6 were designed to change a particular trait of real tern nests, while keeping Field work the other traits fixed. This allowed a comparison between pairs of treatments (referred to as contrasts C1–C5, here The study was conducted at the James Ross Island, NE after; for definitions see the Appendix 1 and Fig. 1) to evalu- Antarctic Peninsula (JRI; within ca 13 km of the Johan ate the effect of each particular trait on predation rate. Gregor Mendel Station: 63°48′S, 57°53′W) during two To assess the effect of parental activity (C3), the T2 nests austral summers – from 1 January to 3 February 2009 and ideally should be compared with same artificial nests located from 3 to 27 January 2011. We mapped the distribution of within active tern colonies. However, as we tried to mini- all breeding birds over ca 117 km2 of ice-free area of the mize disturbance to active tern nests, we did not place artifi- Ulu Peninsula (for environmental characteristics see Láska cial nests within active tern colonies. Hence, we compared et al. 2011). About 340/250 pairs of terns and 38/54 pairs artificial T2 nests directly with the active T1 nests. This of skuas were breeding in the area during the two study procedure violates the rule of fixing all but one variable, seasons; no penguin colonies are known from this part of because these treatments differed not only in the presence of JRI (Weidinger and Pavel in press). Although active tern adult birds in the colony, but also in egg type. Hence we nests in all stages were present throughout the study period, inferred the net effect of parental activity from the difference many tern colonies were already abandoned before our between the composite effect of activity and egg type (C3) arrival. Hence we carefully searched for signs of past nesting and the net effect of egg type (C5). in potential breeding sites. The positions of all current- Both active and artificial nests were inspected in 2–10 d year nests were recorded by GPS and the numbers of adult intervals, depending on site accessibility and weather condi- birds and their activity associated with breeding was recorded tions. An active nest was classified as successful (hatched) throughout the fieldwork. when a chick was found in or close to the nest, or when Active tern nests represented one treatment group in faeces around the nest and adult behavior indicated the an artificial nest experiments. Artificial nests were baited presence of a chick(s). A nest was classified as depredated if

90 0.20 C3 C4 C1 C2

0.15 C5

0.10

Daily predation rate 0.05

0.00 T1 T2 T3 T4 T5 T6 Activity: yes no no no no no Colony: in in in edge edge out Nest: yes yes no no no no Egg: tern quail quail tern quail quail

Figure 1. Daily predation rate (with 95% CI) on active Antarctic tern nests (T1) and artificial nests placed in inactive tern colonies (T2, T3),  100 m from inactive colonies (T4, T5) or  1 km from colonies (T6). Treatment means T1–T6 are based on a hierarchical analysis that accounts for individual nest samples within treatments. For a full description of treatments, nest samples and analyzed contrast (C1–C5) see Appendix 1. Estimates obtained after excluding two nest samples with the highest predation rates (sample 6 and 9) are shown by open circles. the egg(s) disappeared before the expected date of hatching, the binary response variable (survived/failed). Intervals of or if we did not find any signs of chick presence. Other uncertain fate or those that failed for reasons other than pre- sources of mortality (intact frozen eggs, flooding, dead dation were excluded from the analysis, i.e. the survival chicks) were treated as censored cases (see below). An artifi- record for that nest was censored by the time of the preced- cial nest was considered depredated if the egg disappeared ing visit. To facilitate interpretation, we converted results between consecutive nest visits. from DSR to daily predation rates (DPR). To assess whether skuas discriminate between familiar Each experimental treatment included 2–6 nest samples (tern) and novel (quail) egg types and, concurrently, to fur- (colonies or transects), and each sample comprised 18–60 ther validate the use of quail eggs in our nest predation nests, while the effective sample size (  number of survived experiment, we performed a simultaneous prey choice test at nest-days  number of nest visit intervals with a predation 10 skua nests in 2011. We placed one tern and one quail event; Shaffer 2004) varied from 27 to 666 per sample egg at a distance of 1.5–2 m from a skua nest within the field (Appendix 1). To account for hierarchical structure of the of view of the incubating bird. Both eggs were separated data and unbalanced sample sizes, we estimated treatment from each other by a distance of 0.5 m and were placed on means while accounting for the sample effect ‘nested’ within the same substrate (flat stones). The entire setup was moni- the treatment effect. Accordingly, we tested differences tored by a mini video camera for  24 h. From the video among treatments as a set of planned contrasts among record we determined the sequence in which the two eggs individual nest samples (Appendix 1). The contrasts C1–C3 were eventually contacted/consumed by the skua. (effects of nest site and parental activity) are between A subset of active tern nests in both years was monitored independent samples, while the contrasts C4 (effect of nest) by mini video cameras to obtain continuous  24 h records and C5 (effect of egg type) are between paired samples of incubation behavior (unpubl.). As a side product, the (Appendix 1). In addition to test results, we present the effect video records showed that terns resumed incubation soon size expressed as the odds ratio (OR) for daily predation ( 3 min) after disturbances by observers, and that none rate. We did not include covariates in the model. The factors of the nine recorded predation events (all by skuas) was tem- considered important a priori were controlled by the experi- porally associated with a preceding observer visit. mental design, and we believe that the potential effects of unknown factors were randomized by spatial and tem Data analysis poral interspersion of treatments among nest samples. We could not evaluate interactive effects because, owing to As the scheduling of nest visits was strongly constrained by logistical constraints, our experiment was not designed to be weather and logistics, we analyzed nest survival in terms of factorial. daily survival rate (DSR). We used the logistic-exposure method (Shaffer 2004) implemented in PROC GENMOD (SAS Inst.), which is an extension of the logistic regression Results allowing for unequal lengths of nest visit intervals. Each interval between successive nest visits was treated as one Daily predation rate varied markedly among the experi observation and the fate of the nest during the interval was mental treatments (Wald c2  152.5, p  0.001; Fig. 1, 5

91 Appendix 1) as well as among samples within treatments C5: Egg (Wald c2  101.4, p  0.001; Appendix 1). DPR on active 14 tern nests was higher in 2009 than in 2011 (Appendix 1), C4: Nest though the limited sample size and the lack of replication precluded any formal analysis. C3: Activity

Predation on artificial nests was remarkably higher in Contrasts C2: Site inactive colonies than in unoccupied sites far from a colony (C1; OR  11.49, CI: 6.56–20.14; Wald c2  72.77, 1 C1: Site p  0.001), and higher than in unoccupied sites close to a colony edge (C2; OR  2.45, CI: 1.39–4.32; Wald c2  9.57, 01 5101520 1 p  0.002). Predation on artificial nests in inactive colonies Odds ratio (daily predation rate) was markedly higher than predation on active tern nests Figure 2. Effect sizes, expressed as the odds ratio, resulting from (C3; OR  6.36, CI: 3.30–12.25; Wald c2  30.59, 1 analyses of contrasts between nest treatments. C1: inactive colonies p  0.001). The comparison of paired samples showed that vs non-breeding sites  1 km from colonies; C2: inactive predation on quail eggs was marginally higher if they were colonies vs non-breeding sites  100 m from colonies; C3: inactive put in real nests rather than on bare ground (C4; OR  1.61, colonies vs active colonies; C4: real nests vs bare ground at CI: 1.01–2.55; Wald c2  4.06, p  0.044), and that pre paired random sites; C5: quail eggs vs tern eggs. See Fig. 1 for 1 dation tended to be marginally higher on quail eggs than on treatment means and Appendix 1 for descriptions of nest samples real tern eggs, though the difference was far from signifi- and definitions of contrasts. Estimates obtained after excluding the 2 two nest samples with the highest predation rates (sample 6 and 9) cant (C5; OR  1.24, CI: 0.71–2.16; Wald c  0.55, 1 are shown by open circles. p  0.458). Reanalysis after excluding two ‘outlying’ nest samples with the highest DPR (Appendix 1: samples 6 and 9) revealed effects of the same direction, but of smaller size concentrate their foraging activity to currently active (Fig. 2); the difference in DPR between a colony and its edge (disclosed by parental activity) and past (memorized) tern was no longer significant (C2; OR  1.25, CI: 0.68–2.31; breeding sites. While it seems that active tern nests ulti- Wald c2  0.53, p  0.468). mately cannot avoid detection by skuas, changing nesting 1 Of the 10 replicates of prey choice test, skuas either sites (Parmelee 1992, Peter et al. 2008) still may provide consumed (9 cases), or ignored (1 case) both experimental a survival advantage, if skuas divide their nest-searching eggs. If the eggs were consumed, neither the sequence of effort between active and past colonies, and/or if skuas first contact with an egg (5  tern vs 4  quail) nor the switch to active colonies after some delay. sequence of egg consumption (6  tern vs 3  quail) sug- We found strong support for an effect of parental colony gested a clear-cut discrimination between egg types (note defense – predation on artificial nests in abandoned that for the given sample size, the inequality would have colonies was markedly higher than predation on active tern to be at least 8 vs 1 to be significant at p  0.05). nests (Fig. 2: C3). While nests in inactive (not defended) colonies were highly vulnerable (Fig. 1: T2 and T3), parental defense could reduce predation rate in active colonies to Discussion the level observed in artificial nesting sites away from colonies (Fig. 1: T1 vs T6). This suggests that parental defense We found that predation rate by skuas varied among can balance the cost of predator attraction to active colo- locations according to their past occupation by nesting nies. Although the anti-predatory significance of nest posi- terns. Predation on artificial nests was much higher in inac- tion within a colony is well documented in several tern tive tern colonies than in similar sites outside colonies, and species (Becker 1995, Brunton 1997, Hernandez-Matias this effect increased with distance from the colony (Fig. 2: et al. 2003, Silva et al. 2010) and seabirds in general C1 and C2). This effect cannot be accounted for by differ- (Wittenberger and Hunt 1985), it remains to be deter- ences in tern activity (breeding terns were absent from either mined whether communal defense by tern colonies may site by the time of the experiment) or local nest density outweigh the potential benefit of inconspicuous solitary (mean  1.4 nest 100 m22 in largest clusters of nests nesting away from traditional colony sites (Kenyon et al. within colonies [Weidinger and Pavel in press] vs ~ 1 artifi- 2007). Although even solitary nesting would probably not cial nest 100 m22 outside colonies). The remarkably high remain undetected in the simple polar habitat, solitary nests predation in past tern colonies (Fig. 1: T2 and T3) suggests may still benefit from being an unattractive prey, in parti that skuas memorized tern breeding sites and continued cular if skuas increase their foraging effort with local prey to revisit them in search for prey, even weeks after their numbers (Müller-Schwarze and Müller-Schwarze 1973, abandonment. In contrast, the very low predation rate in Emslie et al. 1995). sites not previously occupied by terns (Fig. 1: T6) likely Although skuas seem to have good knowledge of current resulted from random encounters with nests, rather than as well as past tern breeding sites, they still have to use from intentional searches for prey. An intermediate pre some other cues to locate individual nests within a colony. dation rate in sites outside but close to past tern colonies When approached by a predator, incubating terns take off (Fig. 1: T4 and T5) may reflect a higher probability of to engage in mobbing, thus leaving their nests unattended. random encounters in sites close to areas of increased The number of rapid (presumably forced) take-offs seen foraging activity. All this suggests that skuas do not search on video records of incubated nests (ca 1.2 take-offs h21; directly for nests over the whole snow-free area, but unpubl.) suggests that attacks are frequent. Skuas must

92 either remember the position of incubating terns before 1992). Nesting terns, either colonial or solitary, probably flushing them, and/or search directly for the eggs. In either cannot avoid being detected, and they cannot defend their case, attacks have to be short (because of mobbing) and nests by attending them. Yet, by temporarily leaving the nest, cryptic nests and eggs would make detection less likely. they can defend it through communal predator mobbing, Antarctic tern nests in the southern parts of the breeding and at the same time, can benefit from the crypsis of unat- range vary from being placed on bare ground or a shallow tended nests and eggs. scrape in soft substrates to mounds made of small stones (Jabłoński 1995, Higgins and Davies 1996, Kaiser 1996). Acknowledgements – This study was supported by the Czech We found that experimental eggs put in real nests of the lat- Geological Survey and by grants from the Ministry of Education ter type suffered from marginally higher predation rate (MSM 6198959212) and Ministry of Environment (VaV SP II compared to eggs placed on bare ground. Because the eggs 1a9/23/07) of the Czech Republic. The work in Antarctica was in both site types were not attended by adults, the differ conducted under permission from the Ministry of Environment; ential predation could not be an effect of skuas’ short- all field procedures comply with the current laws of the Czech term memory of nest position, but rather a result of direct Republic. We would like to thank the crew members of the Johan Gregor Mendel Station for their support and M. Krist for com- searches. This suggests that the minimization of nest size ments on the manuscript. would provide a selective advantage in terms of reduced predation rate, but this is likely constrained by the need for some nest structure keeping the egg(s) in place and protect- References ing them from flooding (Jabłoński 1995, Fargallo et al. 2001). We can only speculate as to why nests seem to be Becker, P. H. 1995. Effects of coloniality on gull predation on considerably larger (i.e. more conspicuous) than may be common tern (Sterna hirundo) chicks. – Colon. Waterbirds necessary for egg protection – the cross-sectional area of a 18: 11–22. single tern egg (ca 11.3 cm2) is only about 5.5% of the Blanco, G. and Bertellotti, M. 2002. Differential predation by area covered by a typical mound of stones (mean  235  mammals and birds: implications for egg-colour polymor- 12 SE cm2, max  490 cm2, n  57 nests). Although we phism in a nomadic breeding seabird. – Biol. J. Linn. Soc. 75: 137–146. cannot distinguish among the range of alternative explana- Brooke, M. D., Keith, D. and Rov, N. 1999. Exploitation of inland- tions (e.g. sexual selection on nest size; Soler et al. 1998, breeding Antarctic petrels by south polar skuas. – Oecologia Fargallo et al. 2001), of note are two side findings related 121: 25–31. to nest size. First, incubating birds exhibited nest-building Brunton, D. H. 1997. Impacts of predators: center nests are less behavior throughout the incubation period, and such successful than edge nests in a large nesting colony of least behavior was most frequent just before/after change-overs terns. – Condor 99: 372–380. between partners (own video-based data; see also Parmelee Casaux, R., Baroni, A., Ramón, A., Favero, M. and Silva, P. 2008. and Maxson 1974, Kaiser 1996). Second, nests containing Aspects of the foraging behaviour of the Antarctic tern Sterna vittata gaini at Harmony Point, South Shetland Islands. two-egg clutches were marginally larger than nests with 2 – Polar Biol. 31: 327–331. one-egg clutches (mean difference  56  33 SE cm ). Clode, D. 1993. Colonially breeding seabirds: predators or prey? Similar to many other ground-nesting birds with incon- – Trend Ecol. Evol. 8: 336–338. spicuous nests (Götmark 1993), the eggs of the Antarctic Emslie, S. D., Karnovsky, N. and Trivelpiece, W. 1995. Avian tern are remarkably cryptic, at least to the human eye. Our predation at penguin colonies on King George Island, finding that predation did not differ markedly between Antarctica. – Wilson Bull. 107: 317–327. quail and tern eggs validates the use of quail eggs as a surro- Fargallo, J. A., De Leon, A. and Potti, J. 2001. Nest-maintenance gate for real tern eggs to measure relative predation rates. The effort and health status in chinstrap penguins, Pygoscelis antarctica: the functional significance of stone-provisioning result of the simultaneous prey choice experiment suggests behaviour. – Behav. Ecol. Sociobiol. 50: 141–150. that the predation rates found in an artificial nest experiment Götmark, F. 1993. Conspicuous nests may select for non-cryptic resulted from egg/nest detectability, rather than from egg eggs: a comparative study of avian families. – Ornis Fenn. discrimination based on coloration (Montevecchi 1976a), 70: 102–105. size (Montevecchi 1976b), or overall prey novelty (Müller- Hernandez-Matias, A., Jover, L. and Ruiz, X. 2003. Predation on Schwarze and Müller-Schwarze 1973). common tern eggs in relation to sub-colony size, nest aggre The effects of nest size and egg type, if real, weretoo gation and breeding synchrony. – Waterbirds 26: 280–289. small to seriously bias or even reverse the effects of parental Higgins, P. J. and Davies, S. J. J. F. 1996. Handbook of Australian, New Zealand and Antarctic birds, Vol. 3, snipe to pigeons. activity and nest site (Fig. 2). As these results were also robust – Oxford Univ. Press. against outlying data samples (Fig. 2), we conclude that our Jabłoński, B. 1995. Distribution, abundance and biology of the focal tests were not biased by the artificial nest experimental Antarctic tern Sterna vittata Gmelin, 1789 on King George design or the data included, and that the effects of parental Island (South Shetland Islands). – Acta Zool. Cracov. 38: activity and nest site reported here are real. 399–460. Altogether, this study has demonstrated the anti- Kaiser, M. 1996. Untersuchungen zur Biologie und Ökologie der predatory value of multiple nesting characteristics in a Antarktisseeschwalbe (Sterna vittata Gmelin, 1789) während simple avian predator–prey system. Under the constraints der Brutzeit auf King George Island, Südshetlandinseln. – PhD thesis, Humboldt Univ., Berlin. imposed by the polar habitat, where terns and skuas are Kaiser, M., Peter, H.-U. and Gebauer, A. 1988. Zum Bruterfolg forced to co-exist in a limited snow-free area, Antarctic terns und einigen Gelegeparametern der Antarktisseeschwalbe, seem to have developed an efficient strategy for breeding in Sterna vittata (Gmelin, 1789) auf King George Island, the presence of their major avian nest predator (Parmelee Südshetlandinseln. – Beitr. Vogelkd. 34: 317–340.

93 Kenyon, J. K., Smith, B. D. and Butler, R. W. 2007. Can redistri- Silva, L., Gatto, A., Garcia, G. and Yorio, P. 2010. Changes in bution of breeding colonies on a landscape mitigate changing the predator–prey interaction between the kelp gull (Larus predation danger? – J. Avian Biol. 38: 541–551. dominicanus) and royal (Thalasseus maximus) and cayenne terns Láska, K., Barták, M., Hájek, J., Prošek, P. and Bohuslavová, O. (Thalasseus sandvicensis euygnathus) in Punta Leon, Argentina. 2011. Climatic and ecological characteristics of deglaciated – Ornitol. Neotrop. 21: 479–487. area of James Ross Island, Antarctica with a special respect Soler, J. J., Moller, A. P. and Soler, M. 1998. Nest building, to vegetation cover. – Czech Polar Rep. 1: 49–62. sexual selection and parental investment. – Evol. Ecol. 12: Montevecchi, W. A. 1976a. Field experiments on the adaptive 427–441. significance of avian eggshell pigmentation. – Behaviour Van Franeker, J. A., Creuwels, J. C. S., Van der Veer, W., Cleland, S. 63: 26–39. and Robertson, G. 2001. Unexpected effects of climate Montevecchi, W. A. 1976b. Egg size and the egg predatory behav- change on the predation of Antarctic petrels. – Antarct. Sci. iour of crows. – Behaviour 62: 304–320. 13: 430–439. Müller-Schwarze, D. and Müller-Schwarze, C. 1973. Differential Varela, S. A. M., Danchin, E. and Wagner, R. H. 2007. Does predation by South Polar skuas in an Adelie penguin rookery. predation select for or against avian coloniality? A compara- – Condor 75: 127–131. tive analysis. – J. Evol. Biol. 20: 1490–1503. Parmelee, D. F. 1992. Antarctic birds. Ecological and behavioral Varpe, O. and Tveraa, T. 2005. Chick survival in relation to approaches. – Univ. of Minnesota Press. nest site: is the Antarctic petrel hiding from its predator? Parmelee, D. F. and Maxson, S. 1974. The Antarctic terns of – Polar Biol. 28: 388–394. Anvers Island. – Living Bird 13: 233–250. Votier, S. C., Bearhop, S., Ratcliffe, N., Phillips, R. A. and Furness, Peter, H.-U., Buesser, C., Mustafa, O. and Pfeiffer, S. 2008. R. W. 2004. Predation by great skuas at a large Shetland Risk assessment for the Fildes Peninsula and Ardley Island and seabird colony. – J. Appl. Ecol. 41: 1117–1128. the development of management plans for designation as Weidinger, K. 1998. Effect of predation by skuas on breeding Antarctic Specially Protected or Managed Areas. – German success of the Cape petrel Daption capense at Nelson Island, Federal Environment Agency, Dessau. Antarctica. – Polar Biol. 20: 170–177. Sachs, J. L., Hughes, C. R., Nuechterlein, G. L. and Buitron, D. Weidinger, K. and Pavel, V. in press. Abundance and breeding 2007. Evolution of coloniality in birds: a test of hypotheses of the Antarctic tern Sterna vittata at the James Ross and with the red-necked grebe (Podiceps grisegena). – Auk 124: Seymour Islands, NE Antarctic Peninsula. – Polar Biol. 628–642. doi: 10.1007/s00300-012-1251-y Sanchez, J. M., Corbacho, C., del Viejo, A. M. and Parejo, D. Whittam, R. M. and Leonard, M. L. 2000. Characteristics of 2004. Colony-site tenacity and egg color crypsis in the predators and offspring influence nest defense by Arctic and gull-billed tern. – Waterbirds 27: 21–30. common terns. – Condor 102: 301–306. Shaffer, T. L. 2004. A unified approach to analyzing nest success. Wittenberger, J. F. and Hunt, G. L. 1985. The adaptive significance – Auk 121: 526–540. of coloniality in birds. – In: Farner, D. S., King, J. R. and Siegel-Causey, D. and Kharitonov, S. P. 1990. The evolution of Parkes, K. C. (eds), Avian biology. Academic Press, pp. 2–78. coloniality. – In: Power, D. M. (ed.), Current ornithology. Young, E. 1994. Skua and penguin: predator and prey. Plenum Press, pp. 285–330. – Cambridge Univ. Press.

94 Appendix 1

Descriptions of experimental treatments, nest samples and comparisons among them in a study of nest predation in the Antarctic tern.

Treatment Activea Colonyb Nestc Eggd Samplee Year nf DPRg 95% CI C1h C2 C3 C4 C5 T1 yes in yes tern 1 2009 43, 331 0.0452 0.0278–0.0725  2 2011 26, 168 0.0167 0.0054–0.0505  T2 no in yes quail 3 2009 60, 611 0.0913 0.0716–0.1156 2 4 2009 55, 176 0.2040 0.1572–0.2603 2 5 2009 21, 189 0.1037 0.0685–0.1539 2 6 2009 18, 27 0.5078 0.3276–0.6859 2  7 2011 26, 155 0.1063 0.0686–0.1611 2  8 2011 25, 161 0.0857 0.0522–0.1373 2  T3 no in no quail 9 2009 20, 48 0.3387 0.2261–0.4731   2 10 2011 26, 159 0.0883 0.0546–0.1396   2 11 2011 25, 225 0.0467 0.0260–0.0823   2 T4 no edge no tern 12 2009 28, 442 0.0396 0.0250–0.0619  13 2011 23, 109 0.0498 0.0225–0.1067  T5 no edge no quail 14 2009 28, 421 0.0503 0.0333–0.0752 2 2 15 2011 24, 100 0.0539 0.0243–0.1150 2 2 T6 no out no quail 16 2009 30, 666 0.0089 0.0039–0.0196 2 17 2011 29, 305 0.0311 0.0168–0.0568 2 18 2011 38, 430 0.0270 0.0153–0.0469 2 19 2011 31, 391 0.0051 0.0012–0.0200 2 20 2011 32, 373 0.0053 0.0013–0.0211 2 ayes  active natural nests, no  artificial nests. bin  within colony, edge  close to colony ( 100 m), out  far from colony ( 1 km). cyes  real nest, no  bare ground: either random site 10 m apart from paired real nest (T3), or sites in transects at 10 m intervals (T4, T5, T6). dtern  real eggs in active nests (T1) or real eggs collected from abandoned nests (T4), quail  Japanese quail eggs. esamples 6–8 were paired with samples 9–11 (representing the same three colonies); samples 12–13 were paired with samples 14–15 (representing the same two transects of artificial nests). fnumber of nests and effective sample size (  number of survived nest-days  number of observation intervals during which a nest was depredated). gdaily predation rate. hcontrasts among nest samples employed in tests of treatment effects: colony (C1, C2), activity (C3), nest (C4), egg (C5).